From 7fcc2fe39fed1cb98a7374a113ff3800e8f9af80 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 25 Aug 2026 08:45:51 +0000 Subject: net: icmp: avoid invalid transport header access in icmp_send tracepoint syzbot reported a WARNING triggered by DEBUG_NET_WARN_ON_ONCE(): WARNING: at skb_transport_header include/linux/skbuff.h:3087 [inline] WARNING: at udp_hdr include/linux/udp.h:23 [inline] WARNING: at do_trace_event_raw_event_icmp_send include/trace/events/icmp.h:30 [inline] WARNING: at trace_event_raw_event_icmp_send+0x48c/0x6ec include/trace/events/icmp.h:11 Call trace: skb_transport_header include/linux/skbuff.h:3087 [inline] udp_hdr include/linux/udp.h:23 [inline] do_trace_event_raw_event_icmp_send include/trace/events/icmp.h:30 [inline] trace_event_raw_event_icmp_send+0x48c/0x6ec include/trace/events/icmp.h:11 __traceiter_icmp_send include/trace/events/icmp.h:11 [inline] __do_trace_icmp_send include/trace/events/icmp.h:11 [inline] trace_icmp_send+0x320/0x49c include/trace/events/icmp.h:11 __icmp_send+0xcfc/0x11d8 net/ipv4/icmp.c:1013 ipv4_send_dest_unreach net/ipv4/route.c:1280 [inline] ipv4_link_failure+0x57c/0x8dc net/ipv4/route.c:1287 dst_link_failure include/net/dst.h:438 [inline] vti_tunnel_xmit+0xe40/0x17a4 net/ipv4/ip_vti.c:307 TP_fast_assign() unconditionally calls udp_hdr(skb) before checking whether the packet is UDP. Furthermore, __icmp_send() can be invoked from paths (e.g., link failures, ARP errors, forwarding, AF_PACKET) where skb->transport_header was never initialized (~0U). Under CONFIG_DEBUG_NET=y, calling skb_transport_header(skb) triggers DEBUG_NET_WARN_ON_ONCE(!skb_transport_header_was_set(skb)). Fix this by: 1. Only parsing transport info when iph->protocol == IPPROTO_UDP. 2. Using skb_header_pointer() at skb_network_offset(skb) + (iph->ihl << 2) to safely fetch the UDP header without assuming transport_header is set. Fixes: db3efdcf70c7 ("net/ipv4: add tracepoint for icmp_send") Reported-by: syzbot+6d2762674103618994b0@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/6a8d5538.91706f20.ef82.0009.GAE@google.com/T/#u Signed-off-by: Eric Dumazet Cc: Peilin He Cc: xu xin Cc: Steven Rostedt Reviewed-by: Jiayuan Chen Reviewed-by: David Ahern Link: https://patch.msgid.link/20260825084551.1562967-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- include/trace/events/icmp.h | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) (limited to 'include') diff --git a/include/trace/events/icmp.h b/include/trace/events/icmp.h index 09ae115099df..6937b778ae54 100644 --- a/include/trace/events/icmp.h +++ b/include/trace/events/icmp.h @@ -27,17 +27,20 @@ TRACE_EVENT(icmp_send, TP_fast_assign( struct iphdr *iph = ip_hdr(skb); - struct udphdr *uh = udp_hdr(skb); - int proto_4 = iph->protocol; + struct udphdr _uh, *uh = NULL; __be32 *p32; __entry->skbaddr = skb; __entry->type = type; __entry->code = code; - if (proto_4 != IPPROTO_UDP || (u8 *)uh < skb->head || - (u8 *)uh + sizeof(struct udphdr) - > skb_tail_pointer(skb)) { + if (iph->protocol == IPPROTO_UDP) + uh = skb_header_pointer(skb, + skb_network_offset(skb) + + (iph->ihl << 2), + sizeof(_uh), &_uh); + + if (!uh) { __entry->sport = 0; __entry->dport = 0; __entry->ulen = 0; -- cgit From 18666c73afe95eeca8707c699b63f96ce3acda42 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 27 Aug 2026 09:59:36 +0000 Subject: tcp: use GFP_ATOMIC in tcp_send_active_reset() tcp_send_active_reset() can be called from contexts where gfp_any() (in tcp_disconnect()) or sk->sk_allocation (in __tcp_close() and mptcp_do_fastclose()) evaluates to GFP_KERNEL, which includes __GFP_FS and __GFP_DIRECT_RECLAIM. Allocating with GFP_KERNEL while holding the socket lock (sk_lock) creates a lockdep dependency: sk_lock -> fs_reclaim This causes false-positive lockdep circular locking warnings with storage subsystems (such as nvme-tcp) that acquire socket locks in block I/O paths and invoke tcp_disconnect() or close sockets upon teardown: set->srcu -> sk_lock -> fs_reclaim -> elevator_lock -> set->srcu Active resets are small RST packet headers that should never enter direct reclaim or block while holding socket locks. Use sk_gfp_mask(sk, GFP_ATOMIC | __GFP_NOWARN) inside tcp_send_active_reset() and remove its priority argument. This preserves __GFP_MEMALLOC access for SOCK_MEMALLOC sockets, suppresses allocation failure warnings, and aligns with other control packet allocations (e.g. tcp_send_fin(), __tcp_send_ack(), tcp_xmit_probe_skb()). Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet Acked-by: Matthieu Baerts (NGI0) Link: https://patch.msgid.link/20260827095936.551524-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- include/net/tcp.h | 3 +-- net/ipv4/tcp.c | 14 ++++++-------- net/ipv4/tcp_output.c | 4 ++-- net/ipv4/tcp_timer.c | 6 +++--- net/mptcp/protocol.c | 3 +-- net/mptcp/protocol.h | 2 +- 6 files changed, 14 insertions(+), 18 deletions(-) (limited to 'include') diff --git a/include/net/tcp.h b/include/net/tcp.h index 670c20876f26..436495ff2271 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -765,8 +765,7 @@ int tcp_fragment(struct sock *sk, enum tcp_queue tcp_queue, void tcp_send_probe0(struct sock *); int tcp_write_wakeup(struct sock *, int mib); void tcp_send_fin(struct sock *sk); -void tcp_send_active_reset(struct sock *sk, gfp_t priority, - enum sk_rst_reason reason); +void tcp_send_active_reset(struct sock *sk, enum sk_rst_reason reason); int tcp_send_synack(struct sock *); void tcp_push_one(struct sock *, unsigned int mss_now); void __tcp_send_ack(struct sock *sk, u32 rcv_nxt, u16 flags); diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index b4237d0e994d..93d723d8c109 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -3182,8 +3182,7 @@ void __tcp_close(struct sock *sk, long timeout) /* Unread data was tossed, zap the connection. */ NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPABORTONCLOSE); tcp_set_state(sk, TCP_CLOSE); - tcp_send_active_reset(sk, sk->sk_allocation, - SK_RST_REASON_TCP_ABORT_ON_CLOSE); + tcp_send_active_reset(sk, SK_RST_REASON_TCP_ABORT_ON_CLOSE); } else if (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime) { /* Check zero linger _after_ checking for unread data. */ sk->sk_prot->disconnect(sk, 0); @@ -3257,7 +3256,7 @@ adjudge_to_death: struct tcp_sock *tp = tcp_sk(sk); if (READ_ONCE(tp->linger2) < 0) { tcp_set_state(sk, TCP_CLOSE); - tcp_send_active_reset(sk, GFP_ATOMIC, + tcp_send_active_reset(sk, SK_RST_REASON_TCP_ABORT_ON_LINGER); __NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPABORTONLINGER); @@ -3276,7 +3275,7 @@ adjudge_to_death: if (sk->sk_state != TCP_CLOSE) { if (tcp_check_oom(sk, 0)) { tcp_set_state(sk, TCP_CLOSE); - tcp_send_active_reset(sk, GFP_ATOMIC, + tcp_send_active_reset(sk, SK_RST_REASON_TCP_ABORT_ON_MEMORY); __NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPABORTONMEMORY); @@ -3377,14 +3376,14 @@ int tcp_disconnect(struct sock *sk, int flags) } else if (unlikely(tp->repair)) { WRITE_ONCE(sk->sk_err, ECONNABORTED); } else if (tcp_need_reset(old_state)) { - tcp_send_active_reset(sk, gfp_any(), SK_RST_REASON_TCP_STATE); + tcp_send_active_reset(sk, SK_RST_REASON_TCP_STATE); WRITE_ONCE(sk->sk_err, ECONNRESET); } else if (tp->snd_nxt != tp->write_seq && (1 << old_state) & (TCPF_CLOSING | TCPF_LAST_ACK)) { /* The last check adjusts for discrepancy of Linux wrt. RFC * states */ - tcp_send_active_reset(sk, gfp_any(), + tcp_send_active_reset(sk, SK_RST_REASON_TCP_DISCONNECT_WITH_DATA); WRITE_ONCE(sk->sk_err, ECONNRESET); } else if (old_state == TCP_SYN_SENT) @@ -5147,8 +5146,7 @@ int tcp_abort(struct sock *sk, int err) bh_lock_sock(sk); if (tcp_need_reset(sk->sk_state)) - tcp_send_active_reset(sk, GFP_ATOMIC, - SK_RST_REASON_TCP_STATE); + tcp_send_active_reset(sk, SK_RST_REASON_TCP_STATE); tcp_done_with_error(sk, err); bh_unlock_sock(sk); diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 6f4dca4a4de9..c5ffffee4349 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -3849,9 +3849,9 @@ void tcp_send_fin(struct sock *sk) * was unread data in the receive queue. This behavior is recommended * by RFC 2525, section 2.17. -DaveM */ -void tcp_send_active_reset(struct sock *sk, gfp_t priority, - enum sk_rst_reason reason) +void tcp_send_active_reset(struct sock *sk, enum sk_rst_reason reason) { + gfp_t priority = sk_gfp_mask(sk, GFP_ATOMIC | __GFP_NOWARN); struct sk_buff *skb; TCP_INC_STATS(sock_net(sk), TCP_MIB_OUTRSTS); diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c index 1038e7ba9c2e..e56eae4bc341 100644 --- a/net/ipv4/tcp_timer.c +++ b/net/ipv4/tcp_timer.c @@ -126,7 +126,7 @@ static int tcp_out_of_resources(struct sock *sk, bool do_reset) (!tp->snd_wnd && !tp->packets_out)) do_reset = true; if (do_reset) - tcp_send_active_reset(sk, GFP_ATOMIC, + tcp_send_active_reset(sk, SK_RST_REASON_TCP_ABORT_ON_MEMORY); tcp_done(sk); __NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPABORTONMEMORY); @@ -809,7 +809,7 @@ static void tcp_keepalive_timer(struct timer_list *t) goto out; } } - tcp_send_active_reset(sk, GFP_ATOMIC, SK_RST_REASON_TCP_STATE); + tcp_send_active_reset(sk, SK_RST_REASON_TCP_STATE); goto death; } @@ -836,7 +836,7 @@ static void tcp_keepalive_timer(struct timer_list *t) icsk->icsk_probes_out > 0) || (user_timeout == 0 && icsk->icsk_probes_out >= keepalive_probes(tp))) { - tcp_send_active_reset(sk, GFP_ATOMIC, + tcp_send_active_reset(sk, SK_RST_REASON_TCP_KEEPALIVE_TIMEOUT); tcp_write_err(sk); goto out; diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index b474d03620a7..e1f08f71cdb1 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -3109,8 +3109,7 @@ static void mptcp_do_fastclose(struct sock *sk) */ inet_csk(ssk)->icsk_ack.rcv_mss = TCP_MIN_MSS; - tcp_send_active_reset(ssk, ssk->sk_allocation, - SK_RST_REASON_TCP_ABORT_ON_CLOSE); + tcp_send_active_reset(ssk, SK_RST_REASON_TCP_ABORT_ON_CLOSE); unlock: release_sock(ssk); } diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index 06a107d4e839..87ccb84e9927 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -690,7 +690,7 @@ mptcp_send_active_reset_reason(struct sock *sk) enum sk_rst_reason reason; reason = sk_rst_convert_mptcp_reason(subflow->reset_reason); - tcp_send_active_reset(sk, GFP_ATOMIC, reason); + tcp_send_active_reset(sk, reason); } /* Made the fwd mem carried by the given skb available to the msk, -- cgit From 28a57fb2c5df4deb42a06e52fd36c14b37aa0034 Mon Sep 17 00:00:00 2001 From: Dong Chenchen Date: Tue, 25 Aug 2026 20:39:09 +0800 Subject: net: iptunnel: fix stale transport header during tunnel decapsulation Syzbot reported a crash in qdisc_pkt_len_segs_init() caused by a stale transport_header offset after tunnel decapsulation. BUG: unable to handle page fault for address: ffffed102091a42e Oops: Oops: 0000 [#1] SMP KASAN NOPTI CPU: 0 UID: 0 PID: 340 Comm: qdisc_uaf_repro Not tainted 7.2.0-rc4-00061-g248951ddc14d #256 PREEMPT(full) Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 RIP: 0010:__asan_load2 qdisc_pkt_len_segs_init (net/core/dev.c:4145) __dev_queue_xmit (net/core/dev.c:4787) br_dev_queue_push_xmit (net/bridge/br_forward.c:53) br_handle_frame_finish (net/bridge/br_input.c:229) br_handle_frame (net/bridge/br_input.c:315) __netif_receive_skb_core.constprop.0 (net/core/dev.c:6099) __netif_receive_skb_list_core (net/core/dev.c:6287) netif_receive_skb_list_internal (net/core/dev.c:6445) napi_complete_done (net/core/dev.c:6813) gro_cell_poll (net/core/gro_cells.c:74) __napi_poll (net/core/dev.c:7735) net_rx_action (net/core/dev.c:7798 net/core/dev.c:7955) handle_softirqs (kernel/softirq.c:622) do_softirq (kernel/softirq.c:523 kernel/softirq.c:510 ) __local_bh_enable_ip (kernel/softirq.c:450) tun_get_user (drivers/net/tun.c:1986 (discriminator 1)) tun_chr_write_iter (drivers/net/tun.c:2032) The issue is completely latent until qdisc read transport header in commit 7fb4c1967011 ("net: pull headers in qdisc_pkt_len_segs_init()"). The crash requires four conditions to line up: 1. The incoming packet is encapsulated and carries GSO metadata. The outer transport header offset is stored in skb->transport_header while the packet is still in the outer tunnel context. 2. The tunnel receiver strips the outer headers. skb->data is advanced to the inner frame, but skb->transport_header is left pointing to the now-removed outer L4 header, so it becomes a negative offset relative to the new data. 3. The inner frame is not delivered to the local IP stack. Instead, it is forwarded at L2 by a bridge or HSR, so ip_rcv_core() never runs and the transport header is not reset to the inner L4 offset. 4. The forwarding path calls __dev_queue_xmit(), which enters qdisc_pkt_len_segs_init(). That function computes the GSO header length from skb_transport_offset(skb). Because the offset is negative, the unsigned cast overflows and pskb_may_pull(skb, hdr_len + sizeof(struct tcphdr)) reads past the end of the skb, triggering a KASAN fault or page fault. The issue specifically requires GSO packets (shinfo->gso_size != 0), which are processed/aggregated through gro_cells. Fix this by clearing transport_header to the ~0U sentinel in gro_cell for all tunnnel driver. GTP does not support GRO/GSO, drop the evil GSO packets in GTP directly. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: syzbot+83181a31faf9455499c5@syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/69de2bee.a00a0220.475f0.0041.GAE@google.com/T/ Suggested-by: Eric Dumazet Signed-off-by: Dong Chenchen Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/20260825123909.1463121-1-dongchenchen2@huawei.com Signed-off-by: Jakub Kicinski --- drivers/net/gtp.c | 5 +++++ include/linux/skbuff.h | 5 +++++ net/core/gro_cells.c | 2 ++ 3 files changed, 12 insertions(+) (limited to 'include') diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c index 298efc76a56b..69fe5717846b 100644 --- a/drivers/net/gtp.c +++ b/drivers/net/gtp.c @@ -318,6 +318,11 @@ static int gtp_inner_proto(struct sk_buff *skb, unsigned int hdrlen, static int gtp_rx(struct pdp_ctx *pctx, struct sk_buff *skb, unsigned int hdrlen, unsigned int role, __u16 inner_proto) { + if (skb_is_gso(skb)) { + netdev_dbg(pctx->dev, "GSO is not supported in GTP\n"); + goto err; + } + if (!gtp_check_ms(skb, pctx, hdrlen, role, inner_proto)) { netdev_dbg(pctx->dev, "No PDP ctx for this MS\n"); return 1; diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 671c13494566..421f6fc45451 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -3082,6 +3082,11 @@ static inline bool skb_transport_header_was_set(const struct sk_buff *skb) return skb->transport_header != (typeof(skb->transport_header))~0U; } +static inline void skb_unset_transport_header(struct sk_buff *skb) +{ + skb->transport_header = (typeof(skb->transport_header))~0U; +} + static inline unsigned char *skb_transport_header(const struct sk_buff *skb) { DEBUG_NET_WARN_ON_ONCE(!skb_transport_header_was_set(skb)); diff --git a/net/core/gro_cells.c b/net/core/gro_cells.c index 1b84385c04bd..d8c0a2867120 100644 --- a/net/core/gro_cells.c +++ b/net/core/gro_cells.c @@ -22,6 +22,8 @@ int gro_cells_receive(struct gro_cells *gcells, struct sk_buff *skb) if (unlikely(!(dev->flags & IFF_UP))) goto drop; + skb_unset_transport_header(skb); + if (!gcells->cells || skb_cloned(skb) || netif_elide_gro(dev)) { res = netif_rx(skb); goto unlock; -- cgit From ac08d183dac0441e41f77bbad50798fe609d90f1 Mon Sep 17 00:00:00 2001 From: Xuanqiang Luo Date: Fri, 28 Aug 2026 09:29:18 +0800 Subject: raw: annotate disconnect-side IPv4 match writers raw_v4_match() reads inet_daddr, inet_rcv_saddr and sk_bound_dev_if locklessly under RCU. Bind and connect writers are annotated, but __udp_disconnect() still clears the same fields using plain stores. Commit 18f116931f52e ("raw: annotate lockless match fields in raw_v4_match()") added the lockless readers and annotated the raw bind and datagram connect writers. Its v4 revision intentionally left the shared disconnect-side IPv4 writers for follow-up cleanup. Complete that follow-up by using WRITE_ONCE() for the disconnect-side stores, including the inet_rcv_saddr reset in inet_reset_saddr(), to pair with the lockless raw socket matcher. Fixes: 0daf07e52709 ("raw: convert raw sockets to RCU") Link: https://lore.kernel.org/netdev/20260716142958.3064224-1-runyu.xiao@seu.edu.cn/ Suggested-by: Runyu Xiao Signed-off-by: Jackie Liu Signed-off-by: Xuanqiang Luo Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/20260828012918.1461-1-xuanqiang.luo@linux.dev Signed-off-by: Jakub Kicinski --- include/net/ip.h | 3 ++- net/ipv4/udp.c | 4 ++-- 2 files changed, 4 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/include/net/ip.h b/include/net/ip.h index a8f57b4f4aa2..6f602df72ee6 100644 --- a/include/net/ip.h +++ b/include/net/ip.h @@ -704,7 +704,8 @@ static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, static __inline__ void inet_reset_saddr(struct sock *sk) { - inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; + inet_sk(sk)->inet_saddr = 0; + WRITE_ONCE(inet_sk(sk)->inet_rcv_saddr, 0); #if IS_ENABLED(CONFIG_IPV6) if (sk->sk_family == PF_INET6) { struct ipv6_pinfo *np = inet6_sk(sk); diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index af9603217444..6ff5670bf6ed 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -2166,10 +2166,10 @@ int __udp_disconnect(struct sock *sk, int flags) */ sk->sk_state = TCP_CLOSE; - inet->inet_daddr = 0; + WRITE_ONCE(inet->inet_daddr, 0); inet->inet_dport = 0; sock_rps_reset_rxhash(sk); - sk->sk_bound_dev_if = 0; + WRITE_ONCE(sk->sk_bound_dev_if, 0); if (!(sk->sk_userlocks & SOCK_BINDADDR_LOCK)) { inet_reset_saddr(sk); if (sk->sk_prot->rehash && -- cgit From dee44f41f206becb41c492899c1996cfd7f82a1b Mon Sep 17 00:00:00 2001 From: Daehyeon Ko <4ncienth@gmail.com> Date: Wed, 26 Aug 2026 09:39:27 +0900 Subject: vsock/virtio: validate packet source for connected sockets virtio_transport_recv_pkt() looks up sockets first by the full source and destination tuple, then by destination only in the bound table. The fallback is needed for listening and connecting sockets, but sockets remain in the bound table after connect(), so it can also return a non-listening socket. The fallback does not validate the source address. In TCP_SYN_SENT, a RESPONSE from an unrelated source can transition the victim socket to TCP_ESTABLISHED while its stored remote address remains unchanged. Subsequent RW packets from that source are delivered through the same destination-only fallback. This was reproduced with capability-empty processes under different UIDs. The attacker discovered the target tuple through unprivileged AF_VSOCK sock_diag and caused the victim socket to read 16 attacker-chosen bytes; the intended peer-side socket read 0 of those 16 bytes. Add vsock_check_source() to validate the transport, source port and source CID against the peer stored in a non-listening socket. The local transport is the CID exception because its packets are generated internally with VMADDR_CID_LOCAL as their source, including connections using CID aliases. Use the helper after lock_sock() in the virtio receive path. Fixes: 06a8fc78367d ("VSOCK: Introduce virtio_vsock_common.ko") Closes: https://lore.kernel.org/netdev/20260813121236.2328599-1-4ncienth@gmail.com/ Cc: stable@vger.kernel.org Suggested-by: Stefano Garzarella Reviewed-by: Bobby Eshleman Signed-off-by: Daehyeon Ko <4ncienth@gmail.com> Link: https://patch.msgid.link/20260826003929.966160-2-4ncienth@gmail.com Signed-off-by: Jakub Kicinski --- include/net/af_vsock.h | 3 +++ net/vmw_vsock/af_vsock.c | 32 ++++++++++++++++++++++++++++++++ net/vmw_vsock/virtio_transport_common.c | 3 ++- 3 files changed, 37 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/net/af_vsock.h b/include/net/af_vsock.h index 3357ee62d10b..5549298c1ec6 100644 --- a/include/net/af_vsock.h +++ b/include/net/af_vsock.h @@ -229,6 +229,9 @@ struct sock *vsock_find_bound_socket_net(struct sockaddr_vm *addr, struct sock *vsock_find_connected_socket_net(struct sockaddr_vm *src, struct sockaddr_vm *dst, struct net *net); +bool vsock_check_source(const struct vsock_sock *vsk, + const struct vsock_transport *transport, + const struct sockaddr_vm *src); void vsock_remove_sock(struct vsock_sock *vsk); void vsock_for_each_connected_socket(struct vsock_transport *transport, void (*fn)(struct sock *sk)); diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c index a33b2a2d381d..f840498b58af 100644 --- a/net/vmw_vsock/af_vsock.c +++ b/net/vmw_vsock/af_vsock.c @@ -438,6 +438,38 @@ struct sock *vsock_find_connected_socket(struct sockaddr_vm *src, } EXPORT_SYMBOL_GPL(vsock_find_connected_socket); +/** + * vsock_check_source - validate a packet source against a socket peer + * @vsk: socket receiving the packet + * @transport: transport receiving the packet + * @src: source address from the packet + * + * Return: true if the packet arrived on the socket's assigned transport and + * its source matches the stored peer. Loopback packets are generated + * internally and always use the local CID as their source, including + * connections using a valid CID alias. + * + * The caller must hold the socket lock and must not call this for listening + * sockets, which accept packets from any source and have no assigned + * transport. + */ +bool vsock_check_source(const struct vsock_sock *vsk, + const struct vsock_transport *transport, + const struct sockaddr_vm *src) +{ + if (vsk->transport != transport) + return false; + + if (src->svm_port != vsk->remote_addr.svm_port) + return false; + + if (src->svm_cid == vsk->remote_addr.svm_cid) + return true; + + return transport->get_local_cid() == VMADDR_CID_LOCAL; +} +EXPORT_SYMBOL_GPL(vsock_check_source); + void vsock_remove_sock(struct vsock_sock *vsk) { /* Transport reassignment must not remove the binding. */ diff --git a/net/vmw_vsock/virtio_transport_common.c b/net/vmw_vsock/virtio_transport_common.c index 88df82364f77..f225f53ed4ba 100644 --- a/net/vmw_vsock/virtio_transport_common.c +++ b/net/vmw_vsock/virtio_transport_common.c @@ -1836,7 +1836,8 @@ void virtio_transport_recv_pkt(struct virtio_transport *t, * lock_sock (note: listener sockets are not assigned to any transport) */ if (sock_flag(sk, SOCK_DONE) || - (sk->sk_state != TCP_LISTEN && vsk->transport != &t->transport)) { + (sk->sk_state != TCP_LISTEN && + !vsock_check_source(vsk, &t->transport, &src))) { (void)virtio_transport_reset_no_sock(t, skb, net); release_sock(sk); sock_put(sk); -- cgit From 2987ee196c88dbde0463dc87d5fb209c684e34a2 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 27 Aug 2026 16:06:56 +0000 Subject: igmp: convert struct ip_sf_list to RCU Commit 23d2b94043ca ("igmp: Add ip_mc_list lock in ip_check_mc_rcu") added spin_lock_bh(&im->lock) to ip_check_mc_rcu() to prevent a use-after-free while iterating im->sources during concurrent deletions. However, ip_check_mc_rcu() is called from RCU read-side critical sections in packet receive and route lookup fast paths (e.g. __mkroute_output(), ip_route_input_rcu(), and __udp4_lib_rcv()). When igmpv3_send_cr() or igmpv3_send_report() holds &pmc->lock and calls add_grec() -> igmpv3_newpack() -> ip_route_output_ports(), an XFRM policy matching a multicast destination triggers xfrm_tmpl_resolve_one() -> xfrm4_get_saddr() -> __mkroute_output() -> ip_check_mc_rcu(). This attempts to acquire &im->lock while &pmc->lock is already held on the same CPU, triggering a lockdep recursive locking warning / deadlock. Fix this by converting IPv4 struct ip_sf_list to RCU, mirroring the IPv6 implementation in net/ipv6/mcast.c: 1. Add struct rcu_head to struct ip_sf_list and annotate sf_next, sources, and tomb as __rcu pointers. 2. Use rcu_assign_pointer() and kfree_rcu() for list updates and deletions. 3. Remove spin_lock_bh(&im->lock) from ip_check_mc_rcu() and traverse im->sources locklessly with for_each_psf_rcu(), reading and writing counter fields with READ_ONCE() and WRITE_ONCE(). Note: RCU conversion of /proc/net/mcfilter will be done in a separate patch. Fixes: 23d2b94043ca ("igmp: Add ip_mc_list lock in ip_check_mc_rcu") Reported-by: syzbot+3d99fb01bcd740f2fc1e@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=3d99fb01bcd740f2fc1e Signed-off-by: Eric Dumazet Reviewed-by: Ido Schimmel Link: https://patch.msgid.link/20260827160656.903003-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- include/linux/igmp.h | 7 +- net/ipv4/igmp.c | 210 ++++++++++++++++++++++++++++++++------------------- 2 files changed, 135 insertions(+), 82 deletions(-) (limited to 'include') diff --git a/include/linux/igmp.h b/include/linux/igmp.h index 3a2d35a9f307..a0cf0398519f 100644 --- a/include/linux/igmp.h +++ b/include/linux/igmp.h @@ -57,20 +57,21 @@ struct ip_mc_socklist { }; struct ip_sf_list { - struct ip_sf_list *sf_next; + struct ip_sf_list __rcu *sf_next; unsigned long sf_count[2]; /* include/exclude counts */ __be32 sf_inaddr; unsigned char sf_gsresp; /* include in g & s response? */ unsigned char sf_oldin; /* change state */ unsigned char sf_crcount; /* retrans. left to send */ + struct rcu_head rcu; }; struct ip_mc_list { struct in_device *interface; __be32 multiaddr; unsigned int sfmode; - struct ip_sf_list *sources; - struct ip_sf_list *tomb; + struct ip_sf_list __rcu *sources; + struct ip_sf_list __rcu *tomb; unsigned long sfcount[2]; union { struct ip_mc_list *next; diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c index b80b8a92f46e..d56355aca797 100644 --- a/net/ipv4/igmp.c +++ b/net/ipv4/igmp.c @@ -188,6 +188,10 @@ static void ip_ma_put(struct ip_mc_list *im) } } +#define pmc_dereference(e, pmc) \ + rcu_dereference_protected(e, lockdep_is_held(&(pmc)->lock) || \ + lockdep_is_held(&(pmc)->interface->mc_tomb_lock)) + #define for_each_pmc_rcu(in_dev, pmc) \ for (pmc = rcu_dereference(in_dev->mc_list); \ pmc != NULL; \ @@ -198,13 +202,28 @@ static void ip_ma_put(struct ip_mc_list *im) pmc != NULL; \ pmc = rtnl_dereference(pmc->next_rcu)) +#define for_each_psf_mclock(pmc, psf) \ + for (psf = pmc_dereference((pmc)->sources, pmc); \ + psf; \ + psf = pmc_dereference(psf->sf_next, pmc)) + +#define for_each_psf_rcu(im, psf) \ + for (psf = rcu_dereference((im)->sources); \ + psf; \ + psf = rcu_dereference(psf->sf_next)) + +#define for_each_psf_tomb(pmc, psf) \ + for (psf = pmc_dereference((pmc)->tomb, pmc); \ + psf; \ + psf = pmc_dereference(psf->sf_next, pmc)) + static void ip_sf_list_clear_all(struct ip_sf_list *psf) { struct ip_sf_list *next; while (psf) { - next = psf->sf_next; - kfree(psf); + next = rcu_dereference_protected(psf->sf_next, 1); + kfree_rcu(psf, rcu); psf = next; } } @@ -349,7 +368,7 @@ igmp_scount(struct ip_mc_list *pmc, int type, int gdeleted, int sdeleted) struct ip_sf_list *psf; int scount = 0; - for (psf = pmc->sources; psf; psf = psf->sf_next) { + for_each_psf_mclock(pmc, psf) { if (!is_in(pmc, psf, type, gdeleted, sdeleted)) continue; scount++; @@ -494,7 +513,8 @@ static struct sk_buff *add_grec(struct sk_buff *skb, struct ip_mc_list *pmc, struct net *net = dev_net(dev); struct igmpv3_report *pih; struct igmpv3_grec *pgr = NULL; - struct ip_sf_list *psf, *psf_next, *psf_prev, **psf_list; + struct ip_sf_list *psf, *psf_next, *psf_prev; + struct ip_sf_list __rcu **psf_list; int scount, stotal, first, isquery, truncate; unsigned int mtu; @@ -517,7 +537,7 @@ static struct sk_buff *add_grec(struct sk_buff *skb, struct ip_mc_list *pmc, psf_list = sdeleted ? &pmc->tomb : &pmc->sources; - if (!*psf_list) + if (!rcu_access_pointer(*psf_list)) goto empty_source; pih = skb ? igmpv3_report_hdr(skb) : NULL; @@ -533,10 +553,12 @@ static struct sk_buff *add_grec(struct sk_buff *skb, struct ip_mc_list *pmc, } first = 1; psf_prev = NULL; - for (psf = *psf_list; psf; psf = psf_next) { + for (psf = pmc_dereference(*psf_list, pmc); + psf; + psf = psf_next) { __be32 *psrc; - psf_next = psf->sf_next; + psf_next = pmc_dereference(psf->sf_next, pmc); if (!is_in(pmc, psf, type, gdeleted, sdeleted)) { psf_prev = psf; @@ -583,10 +605,12 @@ decrease_sf_crcount: psf->sf_crcount--; if ((sdeleted || gdeleted) && psf->sf_crcount == 0) { if (psf_prev) - psf_prev->sf_next = psf->sf_next; + rcu_assign_pointer(psf_prev->sf_next, + psf_next); else - *psf_list = psf->sf_next; - kfree(psf); + rcu_assign_pointer(*psf_list, + psf_next); + kfree_rcu(psf, rcu); continue; } } @@ -655,28 +679,29 @@ static int igmpv3_send_report(struct in_device *in_dev, struct ip_mc_list *pmc) /* * remove zero-count source records from a source filter list */ -static void igmpv3_clear_zeros(struct ip_sf_list **ppsf) +static void igmpv3_clear_zeros(struct ip_sf_list __rcu **ppsf) { struct ip_sf_list *psf_prev, *psf_next, *psf; psf_prev = NULL; - for (psf = *ppsf; psf; psf = psf_next) { - psf_next = psf->sf_next; + for (psf = rcu_dereference_protected(*ppsf, 1); psf; psf = psf_next) { + psf_next = rcu_dereference_protected(psf->sf_next, 1); if (psf->sf_crcount == 0) { if (psf_prev) - psf_prev->sf_next = psf->sf_next; + rcu_assign_pointer(psf_prev->sf_next, psf_next); else - *ppsf = psf->sf_next; - kfree(psf); - } else + rcu_assign_pointer(*ppsf, psf_next); + kfree_rcu(psf, rcu); + } else { psf_prev = psf; + } } } static void kfree_pmc(struct ip_mc_list *pmc) { - ip_sf_list_clear_all(pmc->sources); - ip_sf_list_clear_all(pmc->tomb); + ip_sf_list_clear_all(rcu_dereference_protected(pmc->sources, 1)); + ip_sf_list_clear_all(rcu_dereference_protected(pmc->tomb, 1)); kfree(pmc); } @@ -710,7 +735,8 @@ static void igmpv3_send_cr(struct in_device *in_dev) igmpv3_clear_zeros(&pmc->sources); } } - if (pmc->crcount == 0 && !pmc->tomb && !pmc->sources) { + if (pmc->crcount == 0 && !rcu_access_pointer(pmc->tomb) && + !rcu_access_pointer(pmc->sources)) { if (pmc_prev) pmc_prev->next = pmc_next; else @@ -896,7 +922,7 @@ static int igmp_xmarksources(struct ip_mc_list *pmc, int nsrcs, __be32 *srcs) int i, scount; scount = 0; - for (psf = pmc->sources; psf; psf = psf->sf_next) { + for_each_psf_mclock(pmc, psf) { if (scount == nsrcs) break; for (i = 0; i < nsrcs; i++) { @@ -927,7 +953,7 @@ static int igmp_marksources(struct ip_mc_list *pmc, int nsrcs, __be32 *srcs) /* mark INCLUDE-mode sources */ scount = 0; - for (psf = pmc->sources; psf; psf = psf->sf_next) { + for_each_psf_mclock(pmc, psf) { if (scount == nsrcs) break; for (i = 0; i < nsrcs; i++) @@ -1228,11 +1254,12 @@ static void igmpv3_add_delrec(struct in_device *in_dev, struct ip_mc_list *im, if (pmc->sfmode == MCAST_INCLUDE) { struct ip_sf_list *psf; + for_each_psf_mclock(im, psf) + psf->sf_crcount = pmc->crcount; pmc->tomb = im->tomb; pmc->sources = im->sources; - im->tomb = im->sources = NULL; - for (psf = pmc->sources; psf; psf = psf->sf_next) - psf->sf_crcount = pmc->crcount; + RCU_INIT_POINTER(im->tomb, NULL); + RCU_INIT_POINTER(im->sources, NULL); } spin_unlock_bh(&im->lock); @@ -1271,9 +1298,18 @@ static void igmpv3_del_delrec(struct in_device *in_dev, struct ip_mc_list *im) if (pmc) { im->interface = pmc->interface; if (im->sfmode == MCAST_INCLUDE) { - swap(im->tomb, pmc->tomb); - swap(im->sources, pmc->sources); - for (psf = im->sources; psf; psf = psf->sf_next) + struct ip_sf_list *sources, *tomb; + + tomb = rcu_replace_pointer(im->tomb, + rcu_dereference_protected(pmc->tomb, 1), + lockdep_is_held(&im->lock)); + rcu_assign_pointer(pmc->tomb, tomb); + + sources = rcu_replace_pointer(im->sources, + rcu_dereference_protected(pmc->sources, 1), + lockdep_is_held(&im->lock)); + rcu_assign_pointer(pmc->sources, sources); + for_each_psf_mclock(im, psf) psf->sf_crcount = in_dev->mr_qrv ?: READ_ONCE(net->ipv4.sysctl_igmp_qrv); } else { @@ -1310,8 +1346,8 @@ static void igmpv3_clear_delrec(struct in_device *in_dev) struct ip_sf_list *psf; spin_lock_bh(&pmc->lock); - psf = pmc->tomb; - pmc->tomb = NULL; + psf = pmc_dereference(pmc->tomb, pmc); + RCU_INIT_POINTER(pmc->tomb, NULL); spin_unlock_bh(&pmc->lock); ip_sf_list_clear_all(psf); } @@ -1990,7 +2026,7 @@ static int ip_mc_del1_src(struct ip_mc_list *pmc, int sfmode, int rv = 0; psf_prev = NULL; - for (psf = pmc->sources; psf; psf = psf->sf_next) { + for_each_psf_mclock(pmc, psf) { if (psf->sf_inaddr == *psfsrc) break; psf_prev = psf; @@ -1999,7 +2035,7 @@ static int ip_mc_del1_src(struct ip_mc_list *pmc, int sfmode, /* source filter not found, or count wrong => bug */ return -ESRCH; } - psf->sf_count[sfmode]--; + WRITE_ONCE(psf->sf_count[sfmode], psf->sf_count[sfmode] - 1); if (psf->sf_count[sfmode] == 0) { ip_rt_multicast_event(pmc->interface); } @@ -2011,19 +2047,28 @@ static int ip_mc_del1_src(struct ip_mc_list *pmc, int sfmode, /* no more filters for this source */ if (psf_prev) - psf_prev->sf_next = psf->sf_next; + rcu_assign_pointer(psf_prev->sf_next, + pmc_dereference(psf->sf_next, pmc)); else - pmc->sources = psf->sf_next; + rcu_assign_pointer(pmc->sources, + pmc_dereference(psf->sf_next, pmc)); #ifdef CONFIG_IP_MULTICAST if (psf->sf_oldin && !IGMP_V1_SEEN(in_dev) && !IGMP_V2_SEEN(in_dev)) { - psf->sf_crcount = in_dev->mr_qrv ?: READ_ONCE(net->ipv4.sysctl_igmp_qrv); - psf->sf_next = pmc->tomb; - pmc->tomb = psf; - rv = 1; - } else + struct ip_sf_list *dpsf = kmalloc_obj(*dpsf, GFP_ATOMIC); + + if (dpsf) { + *dpsf = *psf; + dpsf->sf_crcount = in_dev->mr_qrv ?: + READ_ONCE(net->ipv4.sysctl_igmp_qrv); + rcu_assign_pointer(dpsf->sf_next, + pmc_dereference(pmc->tomb, pmc)); + rcu_assign_pointer(pmc->tomb, dpsf); + rv = 1; + } + } #endif - kfree(psf); + kfree_rcu(psf, rcu); } return rv; } @@ -2060,7 +2105,7 @@ static int ip_mc_del_src(struct in_device *in_dev, __be32 *pmca, int sfmode, err = -EINVAL; if (!pmc->sfcount[sfmode]) goto out_unlock; - pmc->sfcount[sfmode]--; + WRITE_ONCE(pmc->sfcount[sfmode], pmc->sfcount[sfmode] - 1); } err = 0; for (i = 0; i < sfcount; i++) { @@ -2083,7 +2128,7 @@ static int ip_mc_del_src(struct in_device *in_dev, __be32 *pmca, int sfmode, #ifdef CONFIG_IP_MULTICAST pmc->crcount = in_dev->mr_qrv ?: READ_ONCE(net->ipv4.sysctl_igmp_qrv); WRITE_ONCE(in_dev->mr_ifc_count, pmc->crcount); - for (psf = pmc->sources; psf; psf = psf->sf_next) + for_each_psf_mclock(pmc, psf) psf->sf_crcount = 0; igmp_ifc_event(pmc->interface); } else if (sf_setstate(pmc) || changerec) { @@ -2104,7 +2149,7 @@ static int ip_mc_add1_src(struct ip_mc_list *pmc, int sfmode, struct ip_sf_list *psf, *psf_prev; psf_prev = NULL; - for (psf = pmc->sources; psf; psf = psf->sf_next) { + for_each_psf_mclock(pmc, psf) { if (psf->sf_inaddr == *psfsrc) break; psf_prev = psf; @@ -2114,12 +2159,12 @@ static int ip_mc_add1_src(struct ip_mc_list *pmc, int sfmode, if (!psf) return -ENOBUFS; psf->sf_inaddr = *psfsrc; - if (psf_prev) { - psf_prev->sf_next = psf; - } else - pmc->sources = psf; + if (psf_prev) + rcu_assign_pointer(psf_prev->sf_next, psf); + else + rcu_assign_pointer(pmc->sources, psf); } - psf->sf_count[sfmode]++; + WRITE_ONCE(psf->sf_count[sfmode], psf->sf_count[sfmode] + 1); if (psf->sf_count[sfmode] == 1) { ip_rt_multicast_event(pmc->interface); } @@ -2132,13 +2177,15 @@ static void sf_markstate(struct ip_mc_list *pmc) struct ip_sf_list *psf; int mca_xcount = pmc->sfcount[MCAST_EXCLUDE]; - for (psf = pmc->sources; psf; psf = psf->sf_next) + for_each_psf_mclock(pmc, psf) { if (pmc->sfcount[MCAST_EXCLUDE]) { psf->sf_oldin = mca_xcount == psf->sf_count[MCAST_EXCLUDE] && !psf->sf_count[MCAST_INCLUDE]; - } else + } else { psf->sf_oldin = psf->sf_count[MCAST_INCLUDE] != 0; + } + } } static int sf_setstate(struct ip_mc_list *pmc) @@ -2149,27 +2196,31 @@ static int sf_setstate(struct ip_mc_list *pmc) int new_in, rv; rv = 0; - for (psf = pmc->sources; psf; psf = psf->sf_next) { + for_each_psf_mclock(pmc, psf) { if (pmc->sfcount[MCAST_EXCLUDE]) { new_in = mca_xcount == psf->sf_count[MCAST_EXCLUDE] && !psf->sf_count[MCAST_INCLUDE]; - } else + } else { new_in = psf->sf_count[MCAST_INCLUDE] != 0; + } if (new_in) { if (!psf->sf_oldin) { struct ip_sf_list *prev = NULL; - for (dpsf = pmc->tomb; dpsf; dpsf = dpsf->sf_next) { + for_each_psf_tomb(pmc, dpsf) { if (dpsf->sf_inaddr == psf->sf_inaddr) break; prev = dpsf; } if (dpsf) { + struct ip_sf_list *dpsf_next; + + dpsf_next = pmc_dereference(dpsf->sf_next, pmc); if (prev) - prev->sf_next = dpsf->sf_next; + rcu_assign_pointer(prev->sf_next, dpsf_next); else - pmc->tomb = dpsf->sf_next; - kfree(dpsf); + rcu_assign_pointer(pmc->tomb, dpsf_next); + kfree_rcu(dpsf, rcu); } psf->sf_crcount = qrv; rv++; @@ -2181,17 +2232,19 @@ static int sf_setstate(struct ip_mc_list *pmc) * add or update "delete" records if an active filter * is now inactive */ - for (dpsf = pmc->tomb; dpsf; dpsf = dpsf->sf_next) + for_each_psf_tomb(pmc, dpsf) { if (dpsf->sf_inaddr == psf->sf_inaddr) break; + } if (!dpsf) { dpsf = kmalloc_obj(*dpsf, GFP_ATOMIC); if (!dpsf) continue; *dpsf = *psf; /* pmc->lock held by callers */ - dpsf->sf_next = pmc->tomb; - pmc->tomb = dpsf; + rcu_assign_pointer(dpsf->sf_next, + pmc_dereference(pmc->tomb, pmc)); + rcu_assign_pointer(pmc->tomb, dpsf); } dpsf->sf_crcount = qrv; rv++; @@ -2231,7 +2284,7 @@ static int ip_mc_add_src(struct in_device *in_dev, __be32 *pmca, int sfmode, #endif isexclude = pmc->sfmode == MCAST_EXCLUDE; if (!delta) - pmc->sfcount[sfmode]++; + WRITE_ONCE(pmc->sfcount[sfmode], pmc->sfcount[sfmode] + 1); err = 0; for (i = 0; i < sfcount; i++) { err = ip_mc_add1_src(pmc, sfmode, &psfsrc[i]); @@ -2242,7 +2295,7 @@ static int ip_mc_add_src(struct in_device *in_dev, __be32 *pmca, int sfmode, int j; if (!delta) - pmc->sfcount[sfmode]--; + WRITE_ONCE(pmc->sfcount[sfmode], pmc->sfcount[sfmode] - 1); for (j = 0; j < i; j++) (void) ip_mc_del1_src(pmc, sfmode, &psfsrc[j]); } else if (isexclude != (pmc->sfcount[MCAST_EXCLUDE] != 0)) { @@ -2262,7 +2315,7 @@ static int ip_mc_add_src(struct in_device *in_dev, __be32 *pmca, int sfmode, pmc->crcount = in_dev->mr_qrv ?: READ_ONCE(net->ipv4.sysctl_igmp_qrv); WRITE_ONCE(in_dev->mr_ifc_count, pmc->crcount); - for (psf = pmc->sources; psf; psf = psf->sf_next) + for_each_psf_mclock(pmc, psf) psf->sf_crcount = 0; igmp_ifc_event(in_dev); } else if (sf_setstate(pmc)) { @@ -2278,13 +2331,13 @@ static void ip_mc_clear_src(struct ip_mc_list *pmc) struct ip_sf_list *tomb, *sources; spin_lock_bh(&pmc->lock); - tomb = pmc->tomb; - pmc->tomb = NULL; - sources = pmc->sources; - pmc->sources = NULL; + tomb = pmc_dereference(pmc->tomb, pmc); + RCU_INIT_POINTER(pmc->tomb, NULL); + sources = pmc_dereference(pmc->sources, pmc); + RCU_INIT_POINTER(pmc->sources, NULL); pmc->sfmode = MCAST_EXCLUDE; - pmc->sfcount[MCAST_INCLUDE] = 0; - pmc->sfcount[MCAST_EXCLUDE] = 1; + WRITE_ONCE(pmc->sfcount[MCAST_INCLUDE], 0); + WRITE_ONCE(pmc->sfcount[MCAST_EXCLUDE], 1); spin_unlock_bh(&pmc->lock); ip_sf_list_clear_all(tomb); @@ -2866,20 +2919,19 @@ int ip_check_mc_rcu(struct in_device *in_dev, __be32 mc_addr, __be32 src_addr, u rv = 1; } else if (im) { if (src_addr) { - spin_lock_bh(&im->lock); - for (psf = im->sources; psf; psf = psf->sf_next) { + for_each_psf_rcu(im, psf) { if (psf->sf_inaddr == src_addr) break; } if (psf) - rv = psf->sf_count[MCAST_INCLUDE] || - psf->sf_count[MCAST_EXCLUDE] != - im->sfcount[MCAST_EXCLUDE]; + rv = READ_ONCE(psf->sf_count[MCAST_INCLUDE]) || + READ_ONCE(psf->sf_count[MCAST_EXCLUDE]) != + READ_ONCE(im->sfcount[MCAST_EXCLUDE]); else - rv = im->sfcount[MCAST_EXCLUDE] != 0; - spin_unlock_bh(&im->lock); - } else + rv = READ_ONCE(im->sfcount[MCAST_EXCLUDE]) != 0; + } else { rv = 1; /* unspecified source; tentatively allow */ + } } return rv; } @@ -3043,7 +3095,7 @@ static inline struct ip_sf_list *igmp_mcf_get_first(struct seq_file *seq) im = rcu_dereference(idev->mc_list); if (likely(im)) { spin_lock_bh(&im->lock); - psf = im->sources; + psf = pmc_dereference(im->sources, im); if (likely(psf)) { state->im = im; state->idev = idev; @@ -3059,7 +3111,7 @@ static struct ip_sf_list *igmp_mcf_get_next(struct seq_file *seq, struct ip_sf_l { struct igmp_mcf_iter_state *state = igmp_mcf_seq_private(seq); - psf = psf->sf_next; + psf = pmc_dereference(psf->sf_next, state->im); while (!psf) { spin_unlock_bh(&state->im->lock); state->im = state->im->next; @@ -3075,7 +3127,7 @@ static struct ip_sf_list *igmp_mcf_get_next(struct seq_file *seq, struct ip_sf_l state->im = rcu_dereference(state->idev->mc_list); } spin_lock_bh(&state->im->lock); - psf = state->im->sources; + psf = pmc_dereference(state->im->sources, state->im); } out: return psf; -- cgit From c073d1b070f171d206b19c98d71739a97f15b3f1 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 28 Aug 2026 08:45:28 +0000 Subject: ipv6: mcast: use copy-on-write RCU updates in ip6_mc_source() pmc->sflist is read locklessly under rcu_read_lock() by inet6_mc_check() during packet reception in the UDP and RAW multicast receive paths. ip6_mc_source() mutated psl->sl_addr and psl->sl_count in-place when adding or removing a source filter. Additionally, when expanding the filter buffer, newpsl was published via rcu_assign_pointer() before writing the new source into the array. Because 16-byte struct in6_addr writes are not atomic and array shifting is not synchronized with RCU readers, concurrent readers in inet6_mc_check() could read torn IPv6 addresses or observe duplicated/missed source entries. Fix this by switching ip6_mc_source() to copy-on-write RCU updates: allocate and fully populate newpsl before publishing it via rcu_assign_pointer(), and reclaim the old filter via kfree_rcu(), matching ip6_mc_msfilter(). Also remove the now unused IP6_SFBLOCK macro. Fixes: 882ba1f73c06 ("mld: convert ipv6_mc_socklist->sflist to RCU") Signed-off-by: Eric Dumazet Cc: Taehee Yoo Reviewed-by: Ido Schimmel Link: https://patch.msgid.link/20260828084531.1826790-3-edumazet@google.com Signed-off-by: Jakub Kicinski --- include/net/if_inet6.h | 2 -- net/ipv6/mcast.c | 98 ++++++++++++++++++++++++++++---------------------- 2 files changed, 56 insertions(+), 44 deletions(-) (limited to 'include') diff --git a/include/net/if_inet6.h b/include/net/if_inet6.h index 238ad3349456..795fb41b45f5 100644 --- a/include/net/if_inet6.h +++ b/include/net/if_inet6.h @@ -88,8 +88,6 @@ struct ip6_sf_socklist { struct in6_addr sl_addr[] __counted_by(sl_max); }; -#define IP6_SFBLOCK 10 /* allocate this many at once */ - struct ipv6_mc_socklist { struct in6_addr addr; int ifindex; diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c index ec7fac511c8d..66f5858e5fea 100644 --- a/net/ipv6/mcast.c +++ b/net/ipv6/mcast.c @@ -355,12 +355,12 @@ int ip6_mc_source(int add, int omode, struct sock *sk, { struct ipv6_pinfo *inet6 = inet6_sk(sk); struct in6_addr *source, *group; + struct ip6_sf_socklist *newpsl, *psl; struct net *net = sock_net(sk); struct ipv6_mc_socklist *pmc; - struct ip6_sf_socklist *psl; struct inet6_dev *idev; int leavegroup = 0; - int i, j, rv; + int i, j; int err; source = &((struct sockaddr_in6 *)&pgsr->gsr_source)->sin6_addr; @@ -409,13 +409,11 @@ int ip6_mc_source(int add, int omode, struct sock *sk, if (!add) { if (!psl) goto done; /* err = -EADDRNOTAVAIL */ - rv = !0; for (i = 0; i < psl->sl_count; i++) { - rv = !ipv6_addr_equal(&psl->sl_addr[i], source); - if (rv == 0) + if (ipv6_addr_equal(&psl->sl_addr[i], source)) break; } - if (rv) /* source not found */ + if (i == psl->sl_count) /* source not found */ goto done; /* err = -EADDRNOTAVAIL */ /* special case - (INCLUDE, empty) == LEAVE_GROUP */ @@ -424,58 +422,74 @@ int ip6_mc_source(int add, int omode, struct sock *sk, goto done; } + atomic_sub(struct_size(psl, sl_addr, psl->sl_max), + &sk->sk_omem_alloc); + + if (psl->sl_count == 1) { + newpsl = NULL; + } else { + newpsl = sock_kmalloc(sk, struct_size(newpsl, sl_addr, + psl->sl_count - 1), + GFP_KERNEL); + if (!newpsl) { + atomic_add(struct_size(psl, sl_addr, psl->sl_max), + &sk->sk_omem_alloc); + err = -ENOBUFS; + goto done; + } + newpsl->sl_max = psl->sl_count - 1; + newpsl->sl_count = psl->sl_count - 1; + for (j = 0; j < i; j++) + newpsl->sl_addr[j] = psl->sl_addr[j]; + for (j = i + 1; j < psl->sl_count; j++) + newpsl->sl_addr[j - 1] = psl->sl_addr[j]; + } + /* update the interface filter */ ip6_mc_del_src(idev, group, omode, 1, source, 1); - for (j = i+1; j < psl->sl_count; j++) - psl->sl_addr[j-1] = psl->sl_addr[j]; - psl->sl_count--; + rcu_assign_pointer(pmc->sflist, newpsl); + kfree_rcu(psl, rcu); err = 0; goto done; } /* else, add a new source to the filter */ - if (psl && psl->sl_count >= sysctl_mld_max_msf) { + if (psl && psl->sl_count >= READ_ONCE(sysctl_mld_max_msf)) { err = -ENOBUFS; goto done; } - if (!psl || psl->sl_count == psl->sl_max) { - struct ip6_sf_socklist *newpsl; - int count = IP6_SFBLOCK; - - if (psl) - count += psl->sl_max; - newpsl = sock_kmalloc(sk, struct_size(newpsl, sl_addr, count), - GFP_KERNEL); - if (!newpsl) { - err = -ENOBUFS; - goto done; - } - newpsl->sl_max = count; - newpsl->sl_count = count - IP6_SFBLOCK; - if (psl) { - for (i = 0; i < psl->sl_count; i++) - newpsl->sl_addr[i] = psl->sl_addr[i]; - atomic_sub(struct_size(psl, sl_addr, psl->sl_max), - &sk->sk_omem_alloc); + if (psl) { + for (i = 0; i < psl->sl_count; i++) { + if (ipv6_addr_equal(&psl->sl_addr[i], source)) + goto done; /* err = -EADDRNOTAVAIL */ } - rcu_assign_pointer(pmc->sflist, newpsl); - kfree_rcu(psl, rcu); - psl = newpsl; } - rv = 1; /* > 0 for insert logic below if sl_count is 0 */ - for (i = 0; i < psl->sl_count; i++) { - rv = !ipv6_addr_equal(&psl->sl_addr[i], source); - if (rv == 0) /* There is an error in the address. */ - goto done; + + i = psl ? psl->sl_count + 1 : 1; + newpsl = sock_kmalloc(sk, struct_size(newpsl, sl_addr, i), + GFP_KERNEL); + if (!newpsl) { + err = -ENOBUFS; + goto done; } - for (j = psl->sl_count-1; j >= i; j--) - psl->sl_addr[j+1] = psl->sl_addr[j]; - psl->sl_addr[i] = *source; - psl->sl_count++; - err = 0; + newpsl->sl_max = i; + newpsl->sl_count = i; + if (psl) { + for (j = 0; j < psl->sl_count; j++) + newpsl->sl_addr[j] = psl->sl_addr[j]; + } + newpsl->sl_addr[i - 1] = *source; + /* update the interface list */ ip6_mc_add_src(idev, group, omode, 1, source, 1); + + if (psl) + atomic_sub(struct_size(psl, sl_addr, psl->sl_max), + &sk->sk_omem_alloc); + rcu_assign_pointer(pmc->sflist, newpsl); + kfree_rcu(psl, rcu); + err = 0; done: mutex_unlock(&idev->mc_lock); in6_dev_put(idev); -- cgit